Combination of ant colony and student psychology based optimization for the multi-depot electric vehicle routing problem with time windows

• Published in: Cluster computing Vol. 28; no. 2; p. 99
• Main Authors: Wei, Xiaoxu, Niu, Chen, Zhao, Lianzheng, Wang, Yongsheng
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.04.2025; Springer Nature B.V
• Subjects: Accuracy; Algorithms; Ant colony optimization; Benchmarks; Computer Communication Networks; Computer Science; Efficiency; Electric vehicles; Energy consumption; Engineering education; Error reduction; Genetic algorithms; Heuristic; Heuristic methods; Integer programming; Machine learning; Operating Systems; Optimization; Processor Architectures; Psychology; Raw materials; Search process; Transport vehicles; Vehicle routing; Windows (intervals); MDVRPTW-EV; SPBO-ACO hybrid metaheuristic algorithm; Perturbation operator; Unmanned electric loader
• ISSN: 1386-7857; 1573-7543
• DOI: 10.1007/s10586-024-04821-9

As electric vehicle technology continues to mature, a growing number of electric engineering and transport vehicles have been deployed in practice, leading to sustained interest in the Multi-Depot Electric Vehicle Routing Problem with Time Windows, this paper proposes the Student Psychology Based Optimization and Ant Colony Optimization hybrid metaheuristic algorithm(SPBO-ACO), which uses the most appropriate algorithm at different stages of the search process. Enable each algorithm to take advantage of its advantages: in the initial stage, ACO algorithm is good at global search, while in the later stage, SPBO’s local search capability is more prominent to solve the path planning problem. The SPBO-ACO algorithm leverages route length classification, strong and weak perturbations, and learning operators to enhance solution exploration.Testing based on standard MDVRPTW benchmark test instances shows high scalability and stability with 25% of results exceeding or approaching optimal solutions in 20 benchmark cases while 85% have errors compared to optimal solutions that do not exceed 10%. When applied to an industrial setting, the algorithm significantly reduces the electric loader’s driving distance during raw material transporting and filling, demonstrating its practical effectiveness.